Oumou Ly: Welcome to The Breakdown. My name is Oumou. I’m a staff fel­low on the Berkan Klein Center’s Assembly Disnformation pro­gram. Our top­ic of dis­cus­sion today is CDA 230, sec­tion 230 of the Communications Decency Act, oth­er­wise known as the twenty-six words that cre­at­ed the Internet. Today I’m joined by Daphne Keller from the Stanford Cyber Policy Center. So, thank you for being with us today Daphne, I appre­ci­ate it. Especially in help­ing us unpack what has turned out to be such a huge and maybe con­se­quen­tial issue for the November elec­tion and cer­tain­ly for tech­nol­o­gy plat­forms and all of us who care and think about dis­in­for­ma­tion real­ly critically. 

One of the first ques­tions I have for you is kind of a basic one. Can you tell us a lit­tle bit about CDA 230 and why it’s referred to as the twenty-six words that start­ed the Internet? 

Daphne Keller: Sure. so first I strong­ly rec­om­mend Jeff Kosseff’s book, which coined that twenty-six words” phrase. It is a great his­to­ry of CDA 230 and it’s very nar­ra­tive, you know. It just sort of explains what was going on in the cas­es and what was going on in Congress. So, it’s not just some­thing for legal nerds and lawyers. It’s a real­ly use­ful reference. 

But maybe just to explain CDA 230’s role, I’ll pull back a lit­tle bit to the big pic­ture of inter­me­di­ary lia­bil­i­ty law in the US gen­er­al­ly. So inter­me­di­ary lia­bil­i­ty law is the law that tells plat­forms what legal respon­si­bil­i­ty they have for the speech and con­tent post­ed by their users. And US law falls into three buck­ets. There’s a big buck­et which is about copy­right, and there the line point is the Digital Millennium Copyright Act, the DMCA. And it has this very chore­o­graphed notice and take­down process. And through Harvard’s Lumen data­base actu­al­ly, there’s been just amaz­ing doc­u­men­ta­tion of how that process is abused, how much erro­neous over­re­moval it leads to, kind of just what hap­pens in that kind of sys­tem. That’s one big bucket. 

The oth­er big buck­et that does­n’t get a lot of atten­tion is fed­er­al crim­i­nal law. There’s no spe­cial immu­ni­ty for plat­forms for fed­er­al crim­i­nal law, and so if what you’re talk­ing about is things like child sex­u­al abuse mate­r­i­al, mate­r­i­al sup­port of terrorism…those things, the reg­u­lar law applies. There is no immu­ni­ty under CDA 230 or any­thing else. 

And then the last big buck­et, the one we’re here to talk about today is CDA 230, which was enact­ed in 1996 as part of a big pack­age of leg­is­la­tion, some of which was sub­se­quent­ly struck down by the Supreme Court, leav­ing CDA 230 stand­ing as the law of the land. And it’s actu­al­ly a real­ly sim­ple law, even though it’s so wide­ly mis­un­der­stood that there’s now a Twitter account, Bad Section 230 Takes, just to retweet all the mis­rep­re­sen­ta­tions of it that come along. 

But what it says is first, plat­forms are not liable for their users’ speech. Again, for the cat­e­go­ry of claims that are cov­ered. So this isn’t about ter­ror­ism, child sex abuse mate­r­i­al, etc. But for things like state law defama­tion claims, plat­forms are not liable for their users’ speech. 

And the sec­ond thing it says is also, plat­forms are not liable for act­ing in good faith to mod­er­ate con­tent, so to enforce their own poli­cies against con­tent they con­sid­er objec­tion­able. And that sec­ond prob­lem was very much part of what Congress was try­ing to accom­plish with this law. They want­ed to make sure that plat­forms could adopt what we now think of as terms of ser­vice or com­mu­ni­ty guide­lines, and could enforce rules against hate­ful speech, or bul­ly­ing, or pornog­ra­phy, or just the the broad range of human behav­ior that most peo­ple don’t want to see on platforms. 

And the key thing that Congress real­ized, because they had expe­ri­ence with a cou­ple of cas­es that hap­pened at the time, was that if you want plat­forms to mod­er­ate, you need to give them both of those immu­ni­ties. You can’t just say, You’re free to mod­er­ate, go do it,” you have to also say, And, if you under­take to mod­er­ate but you miss some­thing and there’s you know, defama­tion still on the plat­form or what­ev­er, the fact that you tried to mod­er­ate won’t be held against you.” 

And this was real­ly impor­tant to Congress because there’d just been a case where a plat­form that tried to mod­er­ate was tagged as act­ing like an edi­tor or a pub­lish­er and there­fore fac­ing poten­tial liability. 

So that’s the core of CDA 230, and I can talk more if it’s help­ful about sort of the things peo­ple get con­fused about like the wide­spread belief that plat­forms are some­how sup­posed to be neu­tral,” which is— 

Ly: Well yeah! Would you please say a few words about that, yes. 

Keller: Yeah. So, I mean Congress had this inten­tion to get plat­forms to mod­er­ate. They did not want them to be neu­tral, they want­ed the opposite. 

Ly: Right. Exactly right.

Keller: Yeah. But I think a lot of peo­ple find it intu­itive to say well it must be that plat­forms have to be neu­tral. And I think that intu­ition comes from a pre-Internet media envi­ron­ment, where kind of every­thing was either a com­mon car­ri­er like a tele­phone just inter­con­nect­ing every­thing and let­ting every­thing flow freely, or it was like NBC News or The New York Times. It was heav­i­ly edit­ed, and the edi­tor clear­ly was respon­si­ble for every­thing that the reporters put in there. And those two mod­els kind of don’t work for the Internet. If we still had just those two mod­els today we would still have only a very tiny num­ber of elites with access to the micro­phone. And every­body else would still not have the abil­i­ty to broad­cast our voic­es on things like Twitter or YouTube, or what­ev­er that we have today. 

And I think that’s not what any­body wants. What peo­ple gen­er­al­ly want is they do want to be able to speak on the Internet with­out plat­form lawyers check­ing every­thing they say before it goes live. We want that. And we also gen­er­al­ly also want plat­forms to mod­er­ate. We want them to take down offen­sive or obnox­ious or hate­ful or dan­ger­ous but legal speech. And so 230 is the law that allows both of those things to hap­pen at once. 

Ly: Daphne, can you talk a lit­tle bit about the two dif­fer­ent types of immu­ni­ty that are out­lined under CDA 230; we call them short­hand (c)(1) and (c)(2)?

Keller: Sure. So, in the super short­hand, (c)(1) is immu­ni­ty for leav­ing con­tent up, and (c)(2) is immu­ni­ty for tak­ing con­tents down. So, most of the lit­i­ga­tion that we’ve seen his­tor­i­cal­ly under the CDA is about (c)(1). Its cases—often you know, real­ly dis­turb­ing cas­es where some­thing ter­ri­ble hap­pened to some­one on the Internet, and speech defam­ing them was left up or speech threat­en­ing them was left up, or they con­tin­ued to face things that were ille­gal. So those are case about (c)(1). If the plat­form leaves that stuff up, are they liable? 

The sec­ond prong, (c)(2), just has­n’t had near­ly as much atten­tion over the years until now. But that’s the one that says plat­forms can choose their own con­tent mod­er­a­tion pol­i­cy. That they’re not liable for choos­ing to take down con­tent they deem objec­tion­able” as long as they are act­ing in good faith.” And that’s the prob­lem, it does have this good faith require­ment. And part of what the exec­u­tive order is try­ing to do is say, Oh, well you have to meet the good faith require­ment to get any of the immu­ni­ties,” you know. If some­one can show that you are not act­ing in good faith, then you lose this much more eco­nom­i­cal­ly con­se­quen­tial immu­ni­ty under (c)(1) for con­tent that’s on your plat­form that’s illegal. 

And sort of the biggest con­cern I think for many peo­ple there is, if this eco­nom­i­cal­ly essen­tial immu­ni­ty is depen­dent on some gov­ern­ment agency deter­min­ing whether you act­ed in good faith, that intro­duces just a ton of room for pol­i­tics, because my idea of what’s good faith won’t be your idea of what’s good faith won’t be Attorney General Barr’s idea of what’s good faith. And so hav­ing some­thing where polit­i­cal appointees in par­tic­u­lar get to decide what con­sti­tutes good faith and then all of your immu­ni­ties hang in the bal­ance? is real­ly fright­en­ing for companies. 

And inter­est­ing­ly, today we see Republicans call­ing for a fair­ness doc­trine for the Internet,” call­ing for a require­ment of good faith or fair­ness in con­tent mod­er­a­tion, but for a gen­er­a­tion it was you know, lit­er­al­ly part of the GOP plat­form every year to oppose the fair­ness doc­trine that was enforced for broad­cast by the FCC. You know, President Reagan said it was uncon­sti­tu­tion­al. This was just like a core con­ser­v­a­tive cri­tique of big gov­ern­ment sup­press­ing speech for…decades, and now it has become their cri­tique, and they’re ask­ing for state reg­u­la­tion of platforms. 

Ly: That is so inter­est­ing to me. Both that and the fact that you know…CDA 230 in so many ways is what allows Donald Trump’s Twitter account to stay up. So it’s real­ly real­ly inter­est­ing that the GOP has decid­ed to rail against it. 

Keller: It’s fascinating.

Ly: So just recent­ly the pres­i­dent signed an exec­u­tive order con­cern­ing CDA 230 pret­ty direct­ly. There was sort of an episode on social media where the pres­i­dent sent out a tweet, it was then labeled by Twitter—fact-checked in way. Can you talk a lit­tle bit about what the exec­u­tive order does? 

Keller: Sure. So I think… I want­ed to start at a super high lev­el with the exec­u­tive order… In the day or so after it came out, I had mul­ti­ple peo­ple from around the world reach out to me and be like, This is like what hap­pened in Venezuela when Chavez start­ed shut­ting down the radio sta­tions.” You know, it’s just sort of…it has this res­i­dence of like—

Ly: It has that feel. Yeah.

Keller: —there’s a polit­i­cal leader try­ing to pun­ish speech plat­forms for their edi­to­r­i­al poli­cies. And that— You know, before you even get into the weeds, that high-level impact of it is real­ly impor­tant to pay atten­tion to. And that is the rea­son why CDT, the Center for Democracy and Technology in DC, has filed a First Amendment case say­ing this whole thing just can’t stand and…we’ll see what hap­pens with that case. 

So then there are also in the exec­u­tive order four oth­er things that might be big deals. So what is that DOJ is instruct­ed to draft leg­is­la­tion to change 230. So, even­tu­al­ly that will come along and pre­sum­ably it will track the very long list of ideas that are in the DOJ report that came out this week. 

A sec­ond is it instructs fed­er­al agen­cies to inter­pret 230 in the way that the exec­u­tive order does, this way that I think is not sup­port­ed by the statute, that takes the good faith” require­ment and kinda…applies it in places it’s not writ­ten in the statute. Nobody’s quite sure what that means. Because there just aren’t that many sit­u­a­tions where fed­er­al agen­cies…care? about 230, but we’ll see what comes out of that. 

A third is that Attorney General Barr of the DOJ is sup­posed to con­vene state attor­neys gen­er­al to look at a long list of com­plaints. And this is like, if you look at it if you’re an Internet pol­i­cy nerd it’s just all the hot-button issues. Sort of like, are fact-checkers biased? Can algo­rith­mic mod­er­a­tion be biased and—well, it can; how can you reg­u­late that? You know, you will rec­og­nize these things if you look at the list. 

And then the fourth one, and and this is one that I think deserves a lot of atten­tion, is that DOJ is sup­posed to review whether plat­forms—par­tic­u­lar plat­forms are prob­lem­at­ic vehi­cles for gov­ern­ment speech due to view­point dis­crim­i­na­tion,” and then based on that look into whether they can car­ry federally-funded ads. I think for most plat­forms the ad dol­lars part is not that big a deal, but being on a fed­er­al gov­ern­ment block list of you know, plat­forms with dis­ap­proved edi­to­r­i­al policies…just like, has this McCarthyist feel. 

Ly: Can you talk a lit­tle bit about the role of CDA in rela­tion to the busi­ness mod­els that the plat­forms run?

Keller: Sure. So, broad­ly speak­ing, the Internet could not exist the way we know it with­out some­thing like CDA 230. And that’s not just about the Facebooks of the world, that’s about every­thing all up and down the tech­ni­cal stack. You know, DNS providers. Cloudflare. Amazon Web Services and oth­er back­end web host­ing. And also tons of lit­tle com­pa­nies, you know. The knit­ting blog that per­mits com­ments, or the farm equip­ment sell­er that has user feed­back. All of those are pos­si­ble because of CDA 230. And if you pull CDA 230 out of the pic­ture, it’s just very hard to imag­ine the coun­ter­fac­tu­al of how American Internet tech­nol­o­gy and com­pa­nies would’ve evolved. They would’ve evolved some­how, you know. And pre­sum­ably the coun­ter­fac­tu­al is we would have some­thing like what the EU has, which boils down to a notice and take­down mod­el for every kind of legal claim? But they… You know, they bare­ly have an Internet econ­o­my for these kinds of com­pa­nies. There’s a rea­son that things devel­oped the way that they did. 

Ly: Yeah. Do you think that there’s any… Maybe not what you think, because I’m sure that we can all agree this is like­ly to be the case. If the lia­bil­i­ty [?] that 230 offers plat­forms is removed, how would that change the way that plat­form approach con­tent moderation?

Keller: Well, I think a lot of lit­tle com­pa­nies would just get out of the busi­ness entire­ly. And so there’s an advo­ca­cy group in DC called Engine, which rep­re­sents star­tups and small com­pa­nies. And they put togeth­er a real­ly inter­est­ing two-pager on the actu­al cost of defend­ing even friv­o­lous claims in a world with CDA 230 and in a world with­out CDA 230, and it’s basi­cal­ly you know, you’re look­ing at 10 to $30,000 in the best-case sce­nario for a case that goes away very very quick­ly, even now. And that’s not a cost that small com­pa­nies want to incur. And investors…you know, there are all these sur­veys of investors say­ing, I don’t want to invest in new plat­forms to chal­lenge today’s incum­bents if they’re in a state of legal uncer­tain­ty where they could be liable for some­thing at any time.” 

So I think you just elim­i­nate a big swath of the parts of… Of both the exist­ing parts of the Internet that pol­i­cy­mak­ers don’t pay any atten­tion to. Like, make the very very vul­ner­a­ble and some of them go away, and that’s trou­bling. And you cre­ate a lot of prob­lems for any new­com­ers who would actu­al­ly chal­lenge today’s incum­bents and try to rival them in seri­ous user-generated con­tent host­ing services. 

For the big plat­forms, you know for Facebook, for YouTube…they’ll sur­vive some­how. You know, they change their busi­ness mod­el. They prob­a­bly… The eas­i­est thing to do is to use our terms of ser­vice to pro­hib­it a whole lot more, and then just like take down a huge swath so you’re not fac­ing much legal risk. 

Ly: It’s hard to imag­ine liv­ing in that kind of a world. 

Keller: It is. It is.

Ly: Yeah. Thank you so much for join­ing me today, Daphne. This was a great and enlight­en­ing con­ver­sa­tion and I’m sure our view­ers will enjoy it. 

Keller: Thank you for hav­ing me. 

Further Reference

Medium post for this episode, with introduction and edited text